Deep Agents v0.7
Deep Agents v0.7 发布,简化了基础框架,在性能相当的情况下将基础输入令牌数减少了 65%。此次更新旨在优化底层架构效率,移除冗余组件以减轻模型负载。该版本延续了系列迭代路径,专注于提升推理效率与资源利用率,为后续功能扩展奠定更轻量级的技术基础。
EVENT DOSSIER
Deep Agents v0.7 has been released, and the managed service has entered the public beta phase.
On August 7, 2026, LangSmith launched the public beta version of Managed Deep Agents. This version supports the deployment of Deep Agents in managed runtimes, offering persistent execution, memory management, sandbox isolation, and production-ready infrastructure. The Deep Agents v0.7 version, released on July 29, 2026, reduced the number of basic input tokens by 65% while simplifying the underlying framework, aiming to optimize the efficiency of the underlying architecture and reduce model load, thus laying a lighter technical foundation for future feature extensions.
DeepAgents Managed Deep Agents 面向生产环境公开测试
Deep Agents v0.7 has been released, and the managed service has entered the public beta phase.
The basic framework has been simplified; the number of basic input tokens has been reduced by 65% while maintaining comparable performance, aiming to optimize the underlying architecture efficiency and reduce model load.
LangSmith has launched the Managed Deep Agents public beta version, supporting persistent execution, memory management, sandbox isolation, and production-ready infrastructure.
Deep Agents v0.7 发布,简化了基础框架,在性能相当的情况下将基础输入令牌数减少了 65%。此次更新旨在优化底层架构效率,移除冗余组件以减轻模型负载。该版本延续了系列迭代路径,专注于提升推理效率与资源利用率,为后续功能扩展奠定更轻量级的技术基础。
LangSmith 已推出 Managed Deep Agents 公共测试版。该服务支持将 Deep Agents 部署于托管运行时,提供持久化执行、内存、沙箱、通道、评估及生产就绪基础设施功能。